Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
1
An Academic and Industrial view
Joanne Cook – Unilever Phillip Martin – University of Manchester
ADVANCED FLUID ENGINEERING FOR DIGITAL MANUFACTURING
2
OUTLINE
Part 1: An Industry View. Joanne Cook 1. Introduction to FMCG innovation process & challenges 2. In silico ambition 3. CAFE4DM WP2 relevance to Unilever. Part 2: An Academic View. Phillip Martin 1. WP1 – simulations. 2. WP2 – structure-property relationships. 3. WP3 – scale up. 4. WP4 – Innovation management and behavioural change
3
High Volume Enduring
Low volume High turnover
• Renovation, optimisation • Cost, speed
• Trend-led • No surprises ….
4
Raw materials
Prototype Product
Manufacture Distribution
Consumer Use
Feedback
TYPICAL PERSONAL CARE PRODUCT INNOVATION CYCLE
New trend
5
OUR RESULTING DIGITAL AGENDA
Product properties and performance models Equipment Performance Calculators
Engineering Calculator
PC Discover + Category
DIGITAL PRODUCT ENGINEERING
Suite of models for in-silico formulation and processing DIGITAL MANUFACTURING
Novel measurements and analysis of process data
Expert knowledge in the hands of all formulators and engineers
Supply Chain benefits
Process data capture and analysis Novel measurements
Clear Links and
some overlap
6
AMBITION: IN-SILICO-FIRST RECIPE DESIGN
Consumer Need /
Marketing Brief
Formulation Selection
Process Selection
Scale-up / roll out
Performance assessment
In-Silico-First validated Recipe
Iterations as required to meet brief
Optimisation
Simulation & optimisation of product innovation – First exemplar = isotropic liquids
Technology Selection
Formulation Design
Process Design
Scale-up
Performance: foam, cleaning, cost…
Physical properties, viscosity, pH…
Mixing speed, cooling rate etc.
Asset performance estimation: e.g. batch cycle time
Virtual Product Design
7
AMBITION: IN-SILICO PRODUCT & PROCESS ‘SCALE-UP’
Expert knowledge in the hands of all product scientists
Laboratory automation
Pilot plant
Factory
• Linking microstructure to rheology
• Predictive scale up from laboratory to pilot plant to factory
• Accurate scale-up and roll-out over
whole supply chain network
• From models to simulations to predictions
• Data generation is the rate-limiting step
Validation
Utilisation
Construction
High Throughput
Product Assembly
High Throughput Performance Assessment
Structured Data
Capture
Mesoscale modelling
Microstructure
8
AMBITION: QUALITY RIGHT FIRST TIME FEED FORWARD CONTROL: NO RETROSPECTIVE BATCH ADJUSTMENTS
RM quality measurement
In-line probe
Sample point
In QC lab or in field
Fibre optic cables (2 off =out & return)
Revised
Recipe
[impurities] model
Road tanker deliveries RM Storage tank
Factory Control System
Salt
Salt
In-line probe
Sample point
9
Potential solution
Validation
Opportunity to optimise manufacturing
• Current processes are developed empirically guided by off-line quality checks.
• Hypothesis that further optimisation would be possible if product microstructure evolution could be tracked.
• Two approaches: 1. Identify off-the shelf measurements and validate 2. Develop new measurements from scratch.
• In-line measurement technique
ü Tracks changes in opacity (could be microstructure change), optical properties unique to each mixing step.
ü Easy to integrate into pilot plant and potentially factory scale.
Tracked parameter vs. process time in pilot plant trials:
ü Reproducible (features and values) ü Chassis specific fingerprints ü PoP for optimising batch cycle time
Conventional product batch cycle time
Optimised reduced batch cycle time
Optimised & conventional, same end point
AMBITION: PROCESSING TO AN END-POINT
Several process steps could be optimised:
10
AMBITION SUMMARY
DIGITAL PRODUCT ENGINEERING
Suite of models for in-silico formulation and processing DIGITAL MANUFACTURING
Novel measurements and analysis of process data
Expert knowledge in the hands of all formulators and engineers
Supply Chain benefits
Clear Links and
some overlap
In silico product recipe design, simulation, optimisation and scale-up
Combining advanced sensor measurements with process analytics to enable cost, efficiency & quality benefits
11
Determine the structure • Simulated properties
(WP1) • Scattering (X-rays or
neutrons) • Diffusion NMR • Rheology (steady shear
and oscillatory)
WP2:STRUCTURE-PROPERTY RELATIONSHIPS
Structure Viscosity
Viscosity prediction available for R&D and SC.
Fulle
r for
mul
atio
ns
Build predictive or correlative models • Machine learning • Group contribution
models
Technical Challenges and Approaches
ACCELERATING MANUFACTURING: We need to understand how PROCESS INFLUENCES MICROSTRUCTURE
• Same formulation but viscosity differs by order of magnitude - Process design cannot be ignored during product design
• How can we predict properties (rheology) and structure through manufacturing process ?
0.01
0.1
1
10
100
0.1 1 10 100
Stress (Pa)
Vis
co
sity o
f 5
% p
rod
uct
(Pa
s)
Modelling challenges New flow physics with new time and length scales Competition between scales leads to different macroscopic responses
Stir a beaker of (Newtonian) water Fluid forms a hollow whirlpool
Fluids may respond to strain rate alone Water (Newtonian), Ketchup (shear thinning)
May also respond to strain WLM mixtures, polymers, drilling muds....
Many (many) different potential models
Stir a beaker of Polyisobutylene in polybutene Fluid climbs up the spoon
Rheological Phenomena in Focus By D.V. Boger, K. Walters
Modelling of complex fluids
Model microstructure Dissipative particulate dynamics (DPD)
Development of continuum approaches Viscoelastic computational fluid dynamics (CFD) model
Integrate DPD with CFD via constitutive equations Quantify micelles growth, scission/recombination and surfactant intermicellar rates and branch-point formation free energy and mobility for the basic fluid and multicomponent mixtures
Challenges: (loss of) scale up
Scale up historically relied on the scale similarity of turbulence Lab scale
Industrial scale
Turbulent eddies look the same regardless of magnification*
Traditional scale up = match Reynolds numbers
The turbulence is Characterised by a Reynolds number
Requirement for accurate, real-time, in-line measurements at pilot and manufacturing scale – current off-line approaches inadequate
Increase capacity by eliminating slow off-line testing
Characterise product batch variability (e,g. between manufacturing sites or due to subtle changes in biosources/local raw materials)
Accelerate engineering design of processes and products through larger more relevant data sets (samples unchanged)
Low Flow High Flow
Process Analytics
Process analytics - challenges
• Measurement in situ/in-line – how to get probes and sensors in to the vessels or
pipes without disturbing the process
• How do we know what to measure?
• Matching accurate off-line approaches
• Moving from correlative approaches to process understanding
• Commercial off-the-shelf instruments often do not exist
To provide a suite of in-line techniques to enable the rapid development and manufacture of new products
In situ spectroscopy
Process analytics - examples
Determination of SLES quality using spectroscopic techniques for adaptive
predictive process control
OO
SO
-O
O
Na+
DFT Calcs predict vib freqs
Proxy measurements of viscosity Near-IR, Mid-IR and Raman in situ spectroscopy
Partial-least squares regression models
Electrical resistance tomography (ERT)
Process Analytics
Low Flow High Flow
Validation of virtual process models
Accurate, real-time, in-line measurements at pilot and manufacturing scale. Sophisticated process analysers and low cost sensors
Measure critical parameters in manufacturing process
Big data
Electrical Resistance Tomography (ERT)
How to promote behavioural change within this new digital environment? How can leaders can use digital information to make better, quicker decisions? How to use data visualisation techniques and data mining to enhance knowledge sharing? How to enhance the adoption of these technologies and facilitate the radical change in organisation and process of innovation within the company which is required for this new digital workflow?
Innovation Management and Behavioural Change
Thank you